text stringlengths 300 320k | source stringlengths 52 154 |
|---|---|
# Sunday July 12 — talks by Scott Garrabrant, Alexflint, alexei, Stuart_Armstrong
This Sunday at 12:00 (PDT), we're running another session of "lightning talks" by curated LessWrong authors (see [here](https://www.lesswrong.com/posts/x46rsCLk6fyAFQsBz/this-sunday-lightning-talks-by-curated-lw-authors) and [here](https... | https://www.lesswrong.com/posts/jRGmNi6CQFQDBvkub/sunday-july-12-talks-by-scott-garrabrant-alexflint-alexei |
# What problem would you like to see Reinforcement Learning applied to?
We often talk about the dangers and challenges of AI and self-improving agents, but I'm curious what you view as potential beneficial applications of AI - if any! As a ML researcher I encounter a lot of positivity and hype in the field, so the ver... | https://www.lesswrong.com/posts/Dry3vRTAiX2JhghMF/what-problem-would-you-like-to-see-reinforcement-learning |
# Partially Stepping Down Isolation
On March 13th our household [started isolating](https://www.jefftk.com/p/stepping-up-isolation). We've relaxed our rules about surfaces a bit since then in response to changing estimates of risk—we've stopped sanitizing our groceries—but otherwise we haven't changed things much: I h... | https://www.lesswrong.com/posts/5yRqio9rRTjCzZ8aj/partially-stepping-down-isolation |
# UML IV: Linear Predictors
(This is the fourth post in a series on Machine Learning based on [this book](https://www.amazon.com/Understanding-Machine-Learning-Theory-Algorithms/dp/1107057132). Click [here](https://www.lesswrong.com/posts/Z29etNPef6yvrPteK/understanding-machine-learning-i) for part one. If you have so... | https://www.lesswrong.com/posts/hg77gpALMEjbTqxf3/uml-iv-linear-predictors |
# How "honest" is GPT-3?
_Edited to add: There is now a [1k bounty for a closely related question](https://www.lesswrong.com/posts/H9knnv8BWGKj6dZim/usd1000-bounty-for-openai-to-show-whether-gpt3-was)._
In [Inaccessible Information](https://www.lesswrong.com/posts/ZyWyAJbedvEgRT2uF/inaccessible-information#II__Where_... | https://www.lesswrong.com/posts/c3RsLTcxrvH4rXpBL/how-honest-is-gpt-3 |
# Nicotinamide riboside vs. SARS-coV-2 update
You may recall my previous rants about nicotinamide riboside vs. SARS-CoV-2. They have been borne out so far... there has been a human trial on NR in Denmark since May 28, the NIH just announced a hamster preclinical yesterday to be done in their BL4, and Brenner lab h... | https://www.lesswrong.com/posts/AhGTmfnwBtF6cJ4wm/nicotinamide-riboside-vs-sars-cov-2-update |
# What does it mean to apply decision theory?
*Based on discussions with Stuart Armstrong and Daniel Kokotajlo.*
There are two conflicting ways of thinking about foundational rationality arguments such as the VNM theorem.
1. As direct arguments for normative principles. The axioms are supposed to be premises which y... | https://www.lesswrong.com/posts/wgdfBtLmByaKYovYe/what-does-it-mean-to-apply-decision-theory |
# Survival in the immoral maze of college
My composition teacher in college told me that in some pottery schools, the teacher holds up your pot, examines it, comments on it, and then smashes it on the floor. They do this for your first 100 pots.
In that spirit, this post's epistemic status is SMASH THIS P... | https://www.lesswrong.com/posts/6NrnyjzRby5MBewSp/survival-in-the-immoral-maze-of-college |
# AI Research Considerations for Human Existential Safety (ARCHES)
Andrew Critch's ([Academian](https://www.lessestwrong.com/users/academian)) and David Krueger's review of 29 AI (existential) safety research directions, each with an illustrative analogy, examples of current work and potential synergies between resear... | https://www.lesswrong.com/posts/QmfjZMr9HxLwHcDQB/ai-research-considerations-for-human-existential-safety |
# The New Frontpage Design & Opening Tag Creation!
We just pushed a new frontpage design that we've been sitting and iterating on for a while, and we also just enabled the ability for everyone to create tags and edit the descriptions of existing tags.
New Frontpage design
--------------------
[A week ago we asked f... | https://www.lesswrong.com/posts/Kr46cTCoo7nkdRhdN/the-new-frontpage-design-and-opening-tag-creation |
# [Resource Request] What are good resources for best practice for creating bureaucracy?
Especially, after just having read The Utopia of Rules: On Technology, Stupidity, and the Secret Joys of Bureaucracy by David Graeber, I feel that most bureaucracy we have is very bad and might create more harm then benefits.
At... | https://www.lesswrong.com/posts/7NrN6Krnehxf7NEHY/resource-request-what-are-good-resources-for-best-practice |
# Covid 7/9: Lies, Damn Lies and Death Rates
Previously: [Covid 7/2: It Could Be Worse](https://thezvi.wordpress.com/2020/07/02/covid-7-2-it-could-be-worse/)
When I posted the link on Twitter, I joked that we already knew this week’s headline.
It turns out we didn’t. Not quite yet. I’m waiting on that one until next... | https://www.lesswrong.com/posts/FR3EkKa4ntZTsahs7/covid-7-9-lies-damn-lies-and-death-rates |
# Arguments against myopic training
_Note that this post has been edited to clarify the difference between explicitly assigning a reward to an action based on its later consequences, versus implicitly reinforcing an action by assigning high reward during later timesteps when its consequences are observed. I'd previous... | https://www.lesswrong.com/posts/GqxuDtZvfgL2bEQ5v/arguments-against-myopic-training |
# Featured Tags: July 2020
The LessWrong team is continuing to develop and validate the Tagging feature we introduced a few months ago. And we just [opened up tag creation to everyone!](https://www.lesswrong.com/posts/Kr46cTCoo7nkdRhdN/the-new-frontpage-design-and-opening-tag-creation) (at least for now)
As inspirati... | https://www.lesswrong.com/posts/KzuBxB8hteDGGorGE/featured-tags-july-2020 |
# Why is the impact penalty time-inconsistent?
I [showed in a previous post](https://www.lesswrong.com/posts/w8QBmgQwb83vDMXoz/dynamic-inconsistency-of-the-inaction-and-initial-state) that impact penalties were time-inconsistent. But why is this? There are two obvious possibilities:
1. The impact penalty is inconsist... | https://www.lesswrong.com/posts/YjqwTepi53MyM4omT/why-is-the-impact-penalty-time-inconsistent |
# How far along are you on the Lesswrong Path?
_How far along are you on the Path?_
====================================
Some say the Path has always existed and there have always been those who have walked it. The Path has many names and there are many notions of what the Path is.
To some the path is called philoso... | https://www.lesswrong.com/posts/CrJhbdvZSDXFgPoDE/how-far-along-are-you-on-the-lesswrong-path |
# Kelly Bet on Everything
Cross-posted, as always, [from Putanumonit](https://putanumonit.com/2020/07/04/kelly-bet-on-everything/).
* * *
It’s a [core staple](https://putanumonit.com/2017/03/12/goddess-spreadsheet/) of Putanumonit to [apply ideas](https://putanumonit.com/2017/11/07/confessions-of-a-slacker/) from [m... | https://www.lesswrong.com/posts/3pjv6uDvY9sqmsnvY/kelly-bet-on-everything |
# Was a terminal degree necessary for inventing Boyle's desiderata?
My composition teacher in college told me that in some pottery schools, the teacher holds up your pot, examines it, comments on it, and then smashes it on the floor. They do this for your first 100 pots.
In that spirit, this post's episte... | https://www.lesswrong.com/posts/8zTgQYn2mZfPNssiK/was-a-terminal-degree-necessary-for-inventing-boyle-s |
# Frankenstein Delenda Est
_\[Not fiction, but still art. Cross-posted from [Grand, Unified, Crazy](https://grandunifiedcrazy.com/2020/07/10/frankenstein-delenda-est/).\]_
I.
I am terrified by the idea that one day, I will look back on my life, and realize that I helped create a monster. That my actions and my decis... | https://www.lesswrong.com/posts/o7ZSxt4QqJeswjwds/frankenstein-delenda-est |
# Mesa-Optimizers vs “Steered Optimizers”
**Abstract**:
The paper [Risks from learned optimization](https://www.lesswrong.com/s/r9tYkB2a8Fp4DN8yB) introduced the term "inner alignment" in the context of a specific class of scenarios, namely a "base optimizer" which searches over a space of “inner” algorithms. If the ... | https://www.lesswrong.com/posts/SJXujr5a2NcoFebr4/mesa-optimizers-vs-steered-optimizers |
# A space of proposals for building safe advanced AI
I liked [Evan’s post on 11 proposals for safe AGI](https://www.alignmentforum.org/posts/fRsjBseRuvRhMPPE5/an-overview-of-11-proposals-for-building-safe-advanced-ai). However, I was a little confused about why he chose these specific proposals; it feels like we could... | https://www.lesswrong.com/posts/S9GxuAEeQomnLkeNt/a-space-of-proposals-for-building-safe-advanced-ai |
# Talk: Key Issues In Near-Term AI Safety Research
I gave a [talk](https://www.youtube.com/watch?v=LHEE_iqzv-8) for the Foresight Institute yesterday, followed by a talk from Dan Elton (NIH) on explainable AI and a panel discussion that included Robert Kirk and Richard Mallah.
While AI Safety researchers are often pr... | https://www.lesswrong.com/posts/yijG7ptfqFBR8w885/talk-key-issues-in-near-term-ai-safety-research |
# Was a PhD necessary to solve outstanding math problems?
My composition teacher in college told me that in some pottery schools, the teacher holds up your pot, examines it, comments on it, and then smashes it on the floor. They do this for your first 100 pots.
In that spirit, this post's epistemic status... | https://www.lesswrong.com/posts/p7uv4Ekhf8utFKBxa/was-a-phd-necessary-to-solve-outstanding-math-problems |
# Your Prioritization is Underspecified
If you were really convinced that the next task on your to do list was the very best thing to advance your goals you'd feel a lot more interested in it. Some part of you believes this to be the case and other parts very obviously don't. Thus, uncertainty about how to prioritize.... | https://www.lesswrong.com/posts/8tGwy2SrMMyvDHEjX/your-prioritization-is-underspecified |
# Mod Note: Tagging Activity is Not Private
We don’t yet have UI for it, so this isn’t obvious (hence the announcement), but it seemed good to clearly communicate that if you tag a post or vote on the tag relevance of a tag-post combination, this information may be viewed by other users and/or the mod team (after we b... | https://www.lesswrong.com/posts/QjJDQspjTwd6CyNSs/mod-note-tagging-activity-is-not-private |
# Maximal Ventilation
At this point, one of the things I'm most concerned about with covid-19 is breathing air that has been [recently breathed by other people](https://www.nature.com/articles/d41586-020-02058-1). Six foot distancing would be pretty good if it quickly settled out of the air, but it's looking more like... | https://www.lesswrong.com/posts/CaoacS9mBEizFZkXZ/maximal-ventilation |
# Meta Programming GPT: A route to Superintelligence?
Imagine typing the following meta-question into GPT-4, a revolutionary new 20 Trillion parameter language model released in 2021:
"_I asked the superintelligence how to cure cancer. The superintelligence responded ___"
How likely are we to get an actual cure for ... | https://www.lesswrong.com/posts/zZLe74DvypRAf7DEQ/meta-programming-gpt-a-route-to-superintelligence |
# Lurking More Before Joining Complex Conversations
*Opinion About Rationalist Parties:*
At a nerdy party, when people are in the middle of discussing a complicated idea, if a new person joins the conversation, I'd prefer if the norm were to \*not\* try to get that person up to speed. Instead, they lurk for awhile, t... | https://www.lesswrong.com/posts/sj4mWbeiX3XDJr8ZR/lurking-more-before-joining-complex-conversations |
# How well can the GPT architecture solve the parity task?
Suppose I give it pairs of strings and ask it to output 1 if the the number of 1s in the string is even and zero if it's odd.
e. g.
0 -> 0
1 -> 1
11 -> 0
101 -> 0
1101-> 1
10101001 -> 0
111000101110 -> 1
How well does it do on this task? What if we ... | https://www.lesswrong.com/posts/2n2Jge6LknQQBj43N/how-well-can-the-gpt-architecture-solve-the-parity-task |
# Covid-19: Analysis of Mortality Data
Data source: [The CDC](https://data.cdc.gov/NCHS/Weekly-Counts-of-Deaths-by-State-and-Select-Causes/muzy-jte6/data), my spreadsheets are [here](https://docs.google.com/spreadsheets/d/17frXKkiNVygLkhFlj9OQE7luYWDL0xeFbdA4JydTy-Y/edit?usp=sharing) and [here](https://docs.google.com... | https://www.lesswrong.com/posts/cgejYQxWjP3So84FQ/covid-19-analysis-of-mortality-data |
# Science eats its young
Let's start by talking about scientific literacy. I'm going to use a weak definition of scientific literacy, one that simply requires familiarity with the Baconian method of inquiry.
I don't want to place an exact number on this issue, but I'd wager the vast majority of the population of "edu... | https://www.lesswrong.com/posts/hy8ro7gixsbSLco9W/science-eats-its-young |
# School Has It Backwards
Life isn't about getting to the right answers; it's about discovering the right questions.
Crossposted from [Curious Human](https://feng.substack.com/p/school-has-it-backwards).
* * *
In school, we're constantly being asked questions and instantly rewarded when we find the right answers. W... | https://www.lesswrong.com/posts/kXwEzKxdNngcAwhzo/school-has-it-backwards |
# What counts as defection?
*Thanks to Michael Dennis for proposing the formal definition; to Andrew Critch for pointing me in this direction; to Abram Demski for proposing non-negative weighting; and to Alex Appel, Scott Emmons, Evan Hubinger, philh, Rohin Shah, and Carroll Wainwright for their feedback and ideas.*
... | https://www.lesswrong.com/posts/8LEPDY36jBYpijrSw/what-counts-as-defection |
# Six economics misconceptions of mine which I've resolved over the last few years
Here are six cases where I was pretty confident in my understanding of the microeconomics of something, but then later found out I was missing an important consideration.
Thanks to Richard Ngo and Tristan Hume for helpful comments.
He... | https://www.lesswrong.com/posts/MgFDzAfCku9MSDLuw/six-economics-misconceptions-of-mine-which-i-ve-resolved |
# In praise of contributing examples, analogies and lingo
Examples, analogies and lingo all make posts easier to understand. So then, I want to encourage people to write more comments like:
* X is an example of what you're describing.
* Y might be a good analogy for this.
* Z could be a good term for that.
I s... | https://www.lesswrong.com/posts/C3DG7B6ihaGCGvtsG/in-praise-of-contributing-examples-analogies-and-lingo |
# Update more slowly!
My experience forecasting has led me to realize that a key mistake that is often made is updating on new data too quickly. This seems backwards, but I think that often the biggest reason that people both under- and over-react to evidence is that they don't consider the evidence clearly enough, an... | https://www.lesswrong.com/posts/pyxuRf3KW5oMbTxYd/update-more-slowly |
# 2020 LessWrong Demographics Survey Results
Thank you to everyone who participated! You can see the [original post here](https://www.lesswrong.com/posts/mf2XKPfT2Brj8mho7/2020-lesswrong-demographics-survey), but the survey is now closed. Here are the results:
Google forms results: [https://docs.google.com/forms/d/e/... | https://www.lesswrong.com/posts/yn4Aw6jejS3SHKuzu/2020-lesswrong-demographics-survey-results |
# Null-boxing Newcomb’s Problem
The leering statue of the unnamed trickster god stared down at him from its sacred alcove, motionless as always.
Maxwell sighed. He had been praying at the base of the statue for the past hour, and like every time previous, he hadn’t felt anything. The priests had promised salvation fro... | https://www.lesswrong.com/posts/fbjNLjNd4zRbY9Wg2/null-boxing-newcomb-s-problem-2 |
# Roll for Sanity
_\[This is very much a personal-diary type post, but it ends up touching on predictive processing and other aspects of how our brains work. Feels related to_ _[Choosing the Zero Point](https://www.lesswrong.com/posts/rMfpnorsMoRwyn4iP/choosing-the-zero-point). Cross-posted from [Grand, Unified, Crazy... | https://www.lesswrong.com/posts/F5wNXwzZJaD98sSnT/roll-for-sanity |
# Book Review: Fooled by Randomness
The book "Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets" by Nassim Nicholas Taleb was published in 2001. On Amazon, the book has a rating of 4.3, rated by 1,124 people, with 62% of reviewers giving 5 stars. On Goodreads, it has a rating of 4.07 from 1,96... | https://www.lesswrong.com/posts/Zwdd4Fx6HBjqeGRQd/book-review-fooled-by-randomness |
# Algorithmic Intent: A Hansonian Generalized Anti-Zombie Principle
> "Why didn't you tell him the truth? Were you afraid?"
>
> "I'm not _afraid_. I _chose_ not to tell him, because I anticipated negative consequences if I did so."
>
> "What do you think 'fear' _is_, exactly?"
The [Generalized Anti-Zombie Principle](... | https://www.lesswrong.com/posts/sXHQ9R5tahiaXEZhR/algorithmic-intent-a-hansonian-generalized-anti-zombie |
# Limits of Current US Prediction Markets (PredictIt Case Study)
_(Disclaimers: I work in the financial industry, though not in a way related to prediction markets. Anything I write here is my opinion and not that of my employer.
This is a US-centric piece based on a case study of PredictIt: as some people have poin... | https://www.lesswrong.com/posts/c3iQryHA4tnAvPZEv/limits-of-current-us-prediction-markets-predictit-case-study |
# Calibration Practice: Retrodictions on Metaculus
*Edit: there is now a cool app for this over here* [*https://www.quantifiedintuitions.org/pastcasting*](https://www.quantifiedintuitions.org/pastcasting)
I've been working on improving my calibration. I practiced on things like the [Credence Game](http://www.acritch.... | https://www.lesswrong.com/posts/BsBh5ieXP5whXbtTK/calibration-practice-retrodictions-on-metaculus |
# Dremeling
In high school, I was on a robotics team (FRC) who, in honesty, were basically incompetent. (Me included.) We were bad at building a robot, bad at programming it, bad at design, but most relevantly for this post bad at _planning ahead_.
The particular instance which added a term to my internal monologue w... | https://www.lesswrong.com/posts/cKBExLeKFmL2jNucr/dremeling |
# The Goldbach conjecture is probably correct; so was Fermat's last theorem
**EDIT**: Added a section on Euler's conjecture.
# The Goldbach conjecture is likely
The Goldbach conjecture is that "every even integer above two is the sum of two primes". For example, $4=2+2$, $6=3+3$, $8=5+3$, and so on.
Though this is ... | https://www.lesswrong.com/posts/iNFZG4d9W848zsgch/the-goldbach-conjecture-is-probably-correct-so-was-fermat-s |
# Mazes and Duality
*(Talk given at* [*an event on Sunday 28th of June*](https://www.lesswrong.com/posts/iZ3AisoiB6qKY6Ctz/sunday-jun-28-more-online-talks-by-curated-authors)*. johnswentworth is responsible for the talk, David Lambert edited the transcript. *
*If you're a curated author and interested in giving a 5-m... | https://www.lesswrong.com/posts/T8qhPyiFPFcSZ2swC/mazes-and-duality |
# Classification of AI alignment research: deconfusion, "good enough" non-superintelligent AI alignment, superintelligent AI alignment
I want to work on technical AI alignment research. I am trying to choose
an AI alignment area to get into. To do that, I want to understand the
big picture of how to make AI go well th... | https://www.lesswrong.com/posts/wopE4nT28ausKGKwt/classification-of-ai-alignment-research-deconfusion-good |
# New paper: AGI Agent Safety by Iteratively Improving the Utility Function
This post is to announce my new paper [AGI Agent Safety by Iteratively Improving the Utility Function](https://arxiv.org/abs/2007.05411). I am also using this post to add some extra background information that is not on the paper. Questions an... | https://www.lesswrong.com/posts/HWRR8YzuM63yZyTPG/new-paper-agi-agent-safety-by-iteratively-improving-the |
# What do we now know about long-term consequences of a COVID-19 infection?
At the beginning there were concerns about permanently reduced lung-function and CFS after some COVID-19 infections. I haven't kept up with the discourse in the last two months. What's the current state of what we now about the long-term effec... | https://www.lesswrong.com/posts/azQ3Qy4ZDuEPfqbnG/what-do-we-now-know-about-long-term-consequences-of-a-covid |
# Divergence causes isolated demands for rigor
Divergence causes isolated demands for rigor
============================================
If you question enough things you can poke whole in any theory, argument, statement, or any other information-transmitting abstraction.
The fully-fledged argument for this is made ... | https://www.lesswrong.com/posts/fedh48pojrfKHGi9w/divergence-causes-isolated-demands-for-rigor |
# Automatically Turning Off Computer at Night
One of my practical life hacks is having my computer turn off automatically at night, periodically, between 11pm and 7am. I'm guessing this gets to me to bed between 30 and 90 minutes earlier each night.
I can interrupt this process if I want to keep working (or screwing... | https://www.lesswrong.com/posts/T6kjHbzgbNiCQweF2/automatically-turning-off-computer-at-night |
# How should AI debate be judged?
\[Epistemic status: thinking out loud. I haven't thought that much about AI debate, and may be missing basic things.\]
Arguments for the correctness of debate and debate-like systems rely on assumptions like "it's easier to point out problems with an argument than it is to craft misl... | https://www.lesswrong.com/posts/m7oGxvouzzeQKiGJH/how-should-ai-debate-be-judged |
# Alignment proposals and complexity classes
In the original “[AI safety via debate](https://arxiv.org/pdf/1805.00899.pdf)” paper, Geoffrey Irving et al. introduced the concept of analyzing different alignment proposals from the perspective of what complexity class they are able to access under optimal play. I think t... | https://www.lesswrong.com/posts/N64THGX7XNCqRtvPG/alignment-proposals-and-complexity-classes |
# [AN #108]: Why we should scrutinize arguments for AI risk
Alignment Newsletter is a weekly publication with recent content relevant to AI alignment around the world. Find all Alignment Newsletter **[resources here](http://rohinshah.com/alignment-newsletter/)**. In particular, you can look through **[this spreadsheet... | https://www.lesswrong.com/posts/T5awG3XQKJtprABsy/an-108-why-we-should-scrutinize-arguments-for-ai-risk |
# [AN #107]: The convergent instrumental subgoals of goal-directed agents
**Newsletter #107**
Alignment Newsletter is a weekly publication with recent content relevant to AI alignment around the world. Find all Alignment Newsletter **[resources here](http://rohinshah.com/alignment-newsletter/)**. In particular, you c... | https://www.lesswrong.com/posts/pjTF49Rnc878jZSAZ/an-107-the-convergent-instrumental-subgoals-of-goal-directed |
# How big of an issue are patent trolls to the average startup?
There are many individual stories about patent trolls being problematic, but I have a hard time estimating how big of an effect their activity happens to be overall. Can someone point me to guide analysis of the problem overall that estimates their impact... | https://www.lesswrong.com/posts/Ma26BypYzCK6CobGz/how-big-of-an-issue-are-patent-trolls-to-the-average-startup |
# Covid 7/16: Becoming the Mask
Previous week: [Covid 7/9: Lies, Damn Lies and Death Rates](https://thezvi.wordpress.com/2020/07/09/covid-7-9-lies-damn-lies-and-death-rates/)
Two weeks ago I predicted a surge in Covid deaths starting on July 2 and picking up on July 7.
On July 6, the rolling 7-day average of Covid d... | https://www.lesswrong.com/posts/cHGgo6pJnfz2sgxuu/covid-7-16-becoming-the-mask |
# Sunday July 19, 1pm (PDT) — talks by Raemon, ricraz, mr-hire, Jameson Quinn
This Sunday at 1pm (PDT), we're running another session of "lightning talks" by curated LessWrong authors (see [here](https://www.lesswrong.com/tag/lesswrong-events) for previous weeks' transcripts).
* Each talk will be 3-5 minutes follow... | https://www.lesswrong.com/posts/eK34tm6iptbxuPgtj/sunday-july-19-1pm-pdt-talks-by-raemon-ricraz-mr-hire |
# Meta-preferences are weird
*(Talk given at* [*an event on Sunday 28th of June*](https://www.lesswrong.com/posts/iZ3AisoiB6qKY6Ctz/sunday-jun-28-more-online-talks-by-curated-authors)*. Charlie Steiner is responsible for the talk, David Lambert edited the transcript. *
*If you're a curated author and interested in gi... | https://www.lesswrong.com/posts/5REnAD496X7czCCqj/meta-preferences-are-weird |
# Environments as a bottleneck in AGI development
Given a training environment or dataset, a training algorithm, an optimiser, and a model class capable of implementing an AGI (with the right parameters), there are two interesting questions we might ask about how conducive that environment is for training an AGI. The ... | https://www.lesswrong.com/posts/vqpEC3MPioHX7bv4t/environments-as-a-bottleneck-in-agi-development |
# Solving Math Problems by Relay
From September to November 2018 we ran an experiment where people did programming in relay. Each player spent ten minutes on a programming problem before passing on their code to a new player who had not seen any of the previous work. We found that people were able to solve some proble... | https://www.lesswrong.com/posts/DWgWbXRfXLGHPgZJM/solving-math-problems-by-relay |
# Telling more rational stories
My composition teacher in college told me that in some pottery schools, the teacher holds up your pot, examines it, comments on it, and then smashes it on the floor. They do this for your first 100 pots.
In that spirit, this post's epistemic status is SMASH THIS POT.
Eliez... | https://www.lesswrong.com/posts/ngoEeB3xFzYiwKy2D/telling-more-rational-stories |
# Anthropomorphizing Humans
Sometimes, we assign desires and emotions to cars. We say that the car wants to drive, or we say that a strange vibration is the car expressing its displeasure at a too-long gap between oil changes. That’s anthropomorphization: we imagine that non-human objects have desires and emotions dri... | https://www.lesswrong.com/posts/LgogZa25oThmxLSAr/anthropomorphizing-humans |
# Lessons on AI Takeover from the conquistadors
*(Talk given at* [*an event on Sunday 28th of June*](https://www.lesswrong.com/posts/iZ3AisoiB6qKY6Ctz/sunday-jun-28-more-online-talks-by-curated-authors)*. Daniel Kokotajlo is responsible for the talk, David Lambert edited the transcript. *
*If you're a curated author ... | https://www.lesswrong.com/posts/kEtgXdjxA4oWjcLFQ/lessons-on-ai-takeover-from-the-conquistadors |
# Calibrate words, not just probabilities
In Phil Tetlock's book Superforecasters, much emphasis is put on making sure that a forecaster's predictions are well calibrated- if a forecaster gives a 70% chance to 100 different events, we should expect that 70 of those events will have happened, and 30 didn't happen. If o... | https://www.lesswrong.com/posts/qRWY8Zo5Fz6iL4e6H/calibrate-words-not-just-probabilities |
# Collection of GPT-3 results
I kept seeing all kinds of crazy reports about people's experiences with GPT-3, so I figured that I'd start collecting them.
* first gwern's [crazy collection](https://www.gwern.net/GPT-3) of all kinds of prompts, with GPT-3 generating poetry, summarizing stories, rewriting things in d... | https://www.lesswrong.com/posts/6Hee7w2paEzHsD6mn/collection-of-gpt-3-results |
# Why is pseudo-alignment "worse" than other ways ML can fail to generalize?
I have only just read the [mesa optimizers paper](https://arxiv.org/pdf/1906.01820.pdf), and I don't understand what it adds to the pre-existing picture that "ML can fail to generalize outside the train distribution and this is bad."
The dis... | https://www.lesswrong.com/posts/TSmgTGaLyhL965jX6/why-is-pseudo-alignment-worse-than-other-ways-ml-can-fail-to |
# Swiss Political System: More than You ever Wanted to Know (I.)
Swiss political system may be best known for its extensive use of referenda. However, others may argue that its most striking feature is the ability to avoid political polarization. In this respect it may be unique among the western nations.
That being ... | https://www.lesswrong.com/posts/x6hpkYyzMG6Bf8T3W/swiss-political-system-more-than-you-ever-wanted-to-know-i |
# Criticism of some popular LW articles
My composition teacher in college told me that in some pottery schools, the teacher holds up your pot, examines it, comments on it, and then smashes it on the floor. They do this for your first 100 pots.
In that spirit, this post's epistemic status is SMASH THIS POT... | https://www.lesswrong.com/posts/vTbRrxKFGk3jmqYqk/criticism-of-some-popular-lw-articles |
# Pulse and Glide Cycling
About 10 years ago I stopped bicycling very much, because my [knees hurt](https://www.jefftk.com/p/what-fixed-my-knees). Since then I've mostly gotten around by walking, running, and public transport. With COVID, however, I've [started bicycling more](https://www.jefftk.com/p/coming-back-to-b... | https://www.lesswrong.com/posts/KHQoSrapKW7JYzgqB/pulse-and-glide-cycling |
# Praise of some popular LW articles
My composition teacher in college told me that in some pottery schools, the teacher holds up your pot, examines it, comments on it, and then smashes it on the floor. They do this for your first 100 pots.
In that spirit, this post's epistemic status is SMASH THIS POT.
... | https://www.lesswrong.com/posts/QgkG4KvoZTouda4jR/praise-of-some-popular-lw-articles |
# "Should Blackmail Be Legal" Hanson/Zvi Debate (Sun July 26th, 3pm PDT)
This weekend's LessWrong event is a debate! Robin Hanson and Zvi Mowshowitz will be debating blackmail this coming Sunday.
The Zoom link is here: [https://us02web.zoom.us/j/85013937335](https://us02web.zoom.us/j/85013937335). ([Facebook event.](... | https://www.lesswrong.com/posts/fvcx6o529zCZ5yN4Q/should-blackmail-be-legal-hanson-zvi-debate-sun-july-26th |
# What Would I Do? Self-prediction in Simple Algorithms
*(This talk was given at* [*a public online event on Sunday July 12th*](https://www.lesswrong.com/tag/lesswrong-event-transcripts)*. Scott Garrabrant is responsible for the talk, Justis Mills edited the transcript. *
*If you're a curated author and interested in... | https://www.lesswrong.com/posts/PiXS9kE4qX68KveCt/what-would-i-do-self-prediction-in-simple-algorithms |
# Operationalizing Interpretability
[Note: This is a quick version of a post that I'll later update with more code + additional citations, but I think the brainstorming here can be useful for others. So I'm sharing just the list of questions for now.]
This blog post is my attempt to get some more clarity on how to ... | https://www.lesswrong.com/posts/TBetMYhEvNDdeX5ji/operationalizing-interpretability |
# Selling real estate: should you overprice or underprice?
*Warning: Written in a hurry, and I am very much not an expert. Very open to corrections.*
*Warning: I am mainly thinking of hot (rising prices, lots of would-be buyers) residential real estate markets within the USA; this post might or might not generalize b... | https://www.lesswrong.com/posts/wbNrGkmLmFjbkznGf/selling-real-estate-should-you-overprice-or-underprice |
# To what extent is GPT-3 capable of reasoning?
ETA 8/19/20: This interview was conducted with AIDungeon's Dragon model in Custom mode. At the time of writing, this mode's first reply was sampled from GPT-2.
Interview #1
------------
Here are excerpts from a [recent "interview"](https://aidungeon.page.link/?link=ht... | https://www.lesswrong.com/posts/L5JSMZQvkBAx9MD5A/to-what-extent-is-gpt-3-capable-of-reasoning |
# Inefficient doesn't mean indifferent, but it might mean wimpy.
My composition teacher in college told me that in some pottery schools, the teacher holds up your pot, examines it, comments on it, and then smashes it on the floor. They do this for your first 100 pots.
In that spirit, this post's epistemic... | https://www.lesswrong.com/posts/Hi4T87no3f755EbsH/inefficient-doesn-t-mean-indifferent-but-it-might-mean-wimpy |
# Learning Values in Practice
*(This talk was given at* [*a public online event on Sunday July 12th*](https://www.lesswrong.com/tag/lesswrong-event-transcripts)*. Stuart Armstrong is responsible for the talk, Justis Mills edited the transcript. *
*If you're a curated author and interested in giving a 5-min talk, whic... | https://www.lesswrong.com/posts/Xy2AYxpWqJWedFfcD/learning-values-in-practice |
# Thiel on Progress and Stagnation
Peter Thiel is one of the most exciting and original thinkers of our era, but many of his opinions are scattered across a range of talks and articles. So Jeremy Nixon and I have put together an organised presentation of his views on progress and stagnation, in his own words. The full... | https://www.lesswrong.com/posts/Xqcorq5EyJBpZcCrN/thiel-on-progress-and-stagnation |
# Parallels Between AI Safety by Debate and Evidence Law
In this post, I highlight some parallels between [AI Safety by Debate](https://openai.com/blog/debate/) (“Debate”) and [evidence law](https://www.law.cornell.edu/rules/fre).
# Evidence law structures high-stakes arguments with human judges.
The prima facie rea... | https://www.lesswrong.com/posts/SrsH2MyyH8MqH9QSr/parallels-between-ai-safety-by-debate-and-evidence-law |
# "Can you keep this confidential? How do you know?"
Pet peeve about privacy: I think people are woefully inadequate at asking, and answering, "Can you keep this confidential?"
Disclosure: I am not inherently great at keeping information private. By default, if a topic came up in conversation, I would accidentally so... | https://www.lesswrong.com/posts/rz73eva3jv267Hy7B/can-you-keep-this-confidential-how-do-you-know |
# Chains, Bottlenecks and Optimization
Consider an idea consisting of a group of strongly connected sub-ideas. If any sub-idea is an error (doesn’t work), then the whole idea is an error (doesn’t work). We can metaphorically model this as a metal chain made of links. How strong is a chain? How hard can you pull on it ... | https://www.lesswrong.com/posts/Ze6PqJK2jnwnhcpnb/chains-bottlenecks-and-optimization |
# Uncalibrated quantum experiments act clasically
_Epistemic status: solid math, uncertain applicability. Would love to hear from someone who knows more physics than I do!_
# Informal Overview
Here's a pretty obvious fact from probability theory:
> Let A and B be two disjoint events, meaning that at most one... | https://www.lesswrong.com/posts/be2WPh7uwH2TNBzGu/uncalibrated-quantum-experiments-act-clasically |
# The "AI Dungeons" Dragon Model is heavily path dependent (testing GPT-3 on ethics)
[AI Dungeons](https://play.aidungeon.io/) is a text-based video game that uses GPT-3 to generate parts of the story. (To try it, click [here](https://play.aidungeon.io/) → New Singleplayer Game → press 6 for "custom" – but note that y... | https://www.lesswrong.com/posts/EKP4HtaHaSiZL4u4N/the-ai-dungeons-dragon-model-is-heavily-path-dependent |
# $1000 bounty for OpenAI to show whether GPT3 was "deliberately" pretending to be stupider than it is
Twitter thread by Eliezer Yudkowsky, with the bounty in bold:
> So I don't want to sound alarms prematurely, here, but we could possibly be looking at the first case of an AI pretending to be stupider than it is. I... | https://www.lesswrong.com/posts/H9knnv8BWGKj6dZim/usd1000-bounty-for-openai-to-show-whether-gpt3-was |
# How good is humanity at coordination?
When EAs look at the history of nuclear weapons, their reactions tend to fall into two camps.
The first camp (which I am inclined towards) is “Man, what a total mess. There were so many near misses, and people involved did such clearly terrible and risky things like setting up ... | https://www.lesswrong.com/posts/y3jDSoTTdBD9Nj3Gx/how-good-is-humanity-at-coordination |
# Alignment As A Bottleneck To Usefulness Of GPT-3
So there’s this thing where GPT-3 is able to do addition, it has the internal model to do addition, but it takes a little poking and prodding to actually get it to do addition. “Few-shot learning”, as [the paper](https://arxiv.org/abs/2005.14165) calls it. Rather than... | https://www.lesswrong.com/posts/BnDF5kejzQLqd5cjH/alignment-as-a-bottleneck-to-usefulness-of-gpt-3 |
# Competition: Amplify Rohin’s Prediction on AGI researchers & Safety Concerns
_EDIT: The competition is now closed, thanks to everyone who participated! Rohin’s posterior distribution is [here](https://elicit.ought.org/builder/rBxYYzM-f), and winners are in [this comment](https://www.lesswrong.com/posts/Azqmzp5JoXJih... | https://www.lesswrong.com/posts/Azqmzp5JoXJihMcr4/competition-amplify-rohin-s-prediction-on-agi-researchers |
# Fresh Bread
Learning to bake bread was a top quarantine activity for many. For too long, I neglected to post the method my wife and I have been using for years. It’s time to fix that.
This recipe is quick and easy, it’s flexible, and it’s absolutely delicious.
This doesn’t mean there isn’t room for other preferenc... | https://www.lesswrong.com/posts/gB44N5rHmu3nBJDPC/fresh-bread |
# [not ongoing] Thoughts on Proportional voting methods
About this post
===============
*This post was a work-in-progress, I (Jameson) updated it over the course of weeks/months, and the reposted it when it got substantial updates. I have now decided to leave it as it is for historical interest, as the next step woul... | https://www.lesswrong.com/posts/Rh8v4ba9w5KTb6onD/not-ongoing-thoughts-on-proportional-voting-methods |
# More Right
Before I came across LessWrong, I felt like I was the only sane person in the world.$^1$ This [excerpt](http://www.hpmor.com/chapter/51) from HPMOR illustrates those feelings well:
> Harry didn't quite know how to describe in words the sense of kinship he felt with Professor Quirrell, except to say that ... | https://www.lesswrong.com/posts/EaqxZHyvZ2hZmhMZZ/more-right |
# Research agenda for AI safety and a better civilization
This agenda is various ideas that occurred to me throughout the years of reading up on AI progress and AI safety. One big idea is that AI is simultaneously software, applied mathematics as well as philosophy and ideology. AI being philosophy and ideology me... | https://www.lesswrong.com/posts/LFkfjxNQwv2jLhfu3/research-agenda-for-ai-safety-and-a-better-civilization |
# The Basic Double Crux pattern
I’ve spent a lot of time developing tools and frameworks for bridging “intractable” disagreements. I’m also the person affiliated with CFAR who has taught Double Crux the most, and done the most work with it.
People often express to me something to the effect,
> The important thing ab... | https://www.lesswrong.com/posts/hNztRARB52Riy36Kz/the-basic-double-crux-pattern |
# If you are signed up for cryonics with life insurance, how much life insurance did you get and over what term?
Hey all,
I'm been cryo-crastinating, but I'm trying to push through and actually get signed up in the next two months.
A next step is that I need to buy life insurance. I'm a little bit at a loss, becaus... | https://www.lesswrong.com/posts/LuTpsuv2Ti7cii8Y6/if-you-are-signed-up-for-cryonics-with-life-insurance-how |
# Swiss Political System: More than You ever Wanted to Know (II.)
[Previous part](http://250bpm.com/blog:161)
In the previous part of this article we've looked in detail at the Swiss system of direct democracy.
It's hard not to conclude that direct democracy is the reason why Swiss society works so well. However, on... | https://www.lesswrong.com/posts/KTnJ4w4grttfJmwo9/swiss-political-system-more-than-you-ever-wanted-to-know-ii |
# Online Meetup: Balaji Srinivasan on "Nation of Emigrants"
Balaji Srinivasan will speak on **"A Nation of Emigrants: How Technology Will Enable Collective Bargaining with Governments"**. See abstract below.
After the talk and Q&A, we will also allow plenty of time for online mingling and discussion.
[Click here to ... | https://www.lesswrong.com/events/FrraEBcYe6kwsXkq5/online-meetup-balaji-srinivasan-on-nation-of-emigrants |
# Online Meetup: Balaji Srinivasan on "Nation of Emigrants"
In our ongoing SlateStarCodex online series, Balaji Srinivasan will speak on **"A Nation of Emigrants: How Technology Will Enable Collective Bargaining with Governments"**. See abstract below.
After the talk and Q&A, we will also allow plenty of time for onl... | https://www.lesswrong.com/posts/sYqGrLRuXXWuuRgAY/online-meetup-balaji-srinivasan-on-nation-of-emigrants-1 |
# [AN #109]: Teaching neural nets to generalize the way humans would
Alignment Newsletter is a weekly publication with recent content relevant to AI alignment around the world. Find all Alignment Newsletter **[resources here](http://rohinshah.com/alignment-newsletter/)**. In particular, you can look through **[this sp... | https://www.lesswrong.com/posts/TWdnCi4kPjTapYjh6/an-109-teaching-neural-nets-to-generalize-the-way-humans |
# Replicating the replication crisis with GPT-3?
I am getting worried that people are having so much fun doing interesting stuff with GPT-3 and AI Dungeon that they're forgetting how easy it is to fool yourself. Maybe we should think about how many different cognitive biases are in play here? Here are some features th... | https://www.lesswrong.com/posts/4jTb9bc6vcm3T2d8a/replicating-the-replication-crisis-with-gpt-3 |
# Weak HCH accesses EXP
_This post is a follow-up to my “[Alignment proposals and complexity classes](https://www.alignmentforum.org/posts/N64THGX7XNCqRtvPG/alignment-proposals-and-complexity-classes)” post. Thanks to Sam Eisenstat for helping with part of the proof here._
[Previously](https://www.alignmentforum.org/... | https://www.lesswrong.com/posts/CtGH3yEoo4mY2taxe/weak-hch-accesses-exp |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.